GIWRF

The intrusion detection system (IDS) has developed into a crucial element of companies that prevent cyber-criminal actions to safeguard the network, sensitive data, and resources. The proposed strategy utilized many machine learning (ML) models for the IDS due to the efficiency of ML algorithms. Defense techniques are typically built around technologies that track and store data in a variety of ways, the most basic of which is the creation of data files, coined security logs. Security analysts typically check such log files in a semi-automatic manner to find events that might compromise the confidentiality, availability, and integrity of the system. Based on this, we suggest an element-free technique to identify security issues in categorized sample data, regardless of their type. The research was conducted on the UNSW-NB15 network dataset to determine the model’s performance. However, the decision tree (DT), gradient boosting tree (GBT), AdaBoost, long short-term memory (LSTM), gated recurrent unit (GRU), and multilayer perceptron (MLP) were trained and tested for the classification in binary tasks to conduct the performance study. Because the effectiveness of IDS degrades with a high vector of dimensional features, a Gini impurity-based weighted RF (GIWRF) model was used as the integrated attribute selection method. This strategy changed the classes of the imbalanced data’s weights and employed the splitting method with the Gini impurity criteria for trees to help the analysis of the class distribution by ML algorithms. The 20 features were chosen from UNSW-NB15 based on their importance. The outcomes of the experiment revealed that DT outperformed the different trained models in this experiment when using the feature selection technique.